FEATURE ARTICLE
Explosives Detection with Nuclear Quadrupole Resonance
An emerging technology will help to uncover land mines and terrorist bombs
Joel Miller, Geoffrey Barrall
Hidden bombs pose an enormous menace. Some are responsible for such
well-publicized atrocities as the downing of two Russian jetliners
on the same day last August. Those in the many millions of land
mines that now litter the globe garner fewer headlines on any
particular day but do far more damage overall, killing or injuring
more than 10,000 people each year. How can civilized society rid
itself of such threats? Clearly, the answer is complicated, but one
component of the solution is to devise equipment that can reliably
uncover concealed explosives before they do harm.
We have been focused on that task for many years now. Specifically,
we and our many government, academic and industry colleagues are
trying to develop the means for detecting explosive chemicals based
on a phenomenon called nuclear quadrupole resonance (NQR). This
approach offers some distinct advantages over the other options
available. For example, the ability of bomb-sniffing dogs and vapor
detectors to sense explosives is influenced by environmental factors
such as wind and ground moisture; also, these approaches can fail
with an explosive that is hermetically sealed, as is the case for
some types of land mines and could readily be arranged in a
terrorist bomb. And one of the most high-tech tactics tried so
far—sensing the nitrogen in explosives using thermal-neutron
analysis—has proved to have inadequate sensitivity and
specificity. Detection through NQR does not face these difficulties.
To understand why not, it is helpful to review the basic physics
behind this promising technique.
Squashed Nuclei
Nuclear quadrupole resonance has much in common with nuclear
magnetic resonance (NMR), the fundamental physical process that
makes magnetic resonance imaging possible. Nuclear magnetic
resonance, first demonstrated in 1946, takes advantage of the fact
that certain atomic nuclei possess magnetic dipole
moments—that is, these nuclei act like tiny bar magnets,
each with a north magnetic pole at one end and a south magnetic pole
at the other. The laws of quantum mechanics dictate that when such
nuclei are subjected to an externally applied magnetic field, they
must align themselves along it. But the magnetic moments of these
nuclei, usually depicted as arrows, are allowed two possible
orientations: in the same direction as the applied magnetic field or
opposite to it.
Although alignment with the applied field is favored (this
being the lower-energy condition), the energy difference between the
two orientations is such that thermal agitation is usually
sufficient to ensure that only slightly more than half the nuclei
are in the lower-energy state. The key is that the nuclei can occupy
two distinct states separated by a well-defined increment in energy.
(It will be well defined as long as the applied magnetic field is
uniform.) In that sense, the situation is much like that of an
electron in an atom, which can be in the "ground" state or
in a higher-energy "excited" state.
A ground-state electron shifts to an excited state when the atom
receives a dollop of electromagnetic radiation of just the right
energy to put it there—that is, when it absorbs a photon of
just the right frequency. Conversely, if this excited-state electron
falls back to the ground state, the atom will emit a photon of the
exact same frequency to carry away the difference in energy. In NMR,
the energy difference between states is much less than for the
electronic states of an atom, so the relevant frequencies are much
lower. Rather than dealing with optical frequencies, NMR typically
involves oscillations of just a few tens to hundreds of megahertz,
which includes the band where broadcast FM radio stations operate.
Nuclear quadrupole resonance is similar in concept, but unlike NMR
it does not rely on the nuclei aligning themselves in an externally
applied magnetic field. Instead, NQR exploits the fact that some
nuclei possess an electric quadrupole moment, which can be
thought of as arising from two back-to-back electric dipoles
(positive and negative charges separated by a short distance). Why
do some atomic nuclei have an electric quadrupole moment? Physicists
would say because they have a spin quantum number greater than
½. A more intuitive explanation is because the positive
electric charge these nuclei carry is not distributed with perfect
spherical symmetry.
Consider for a moment a spherical nucleus with its positive charge
distributed uniformly throughout. Now squeeze that nucleus in your
mind's eye so that what was originally shaped like a basketball is
flattened into a pumpkin. A pumpkin of positive charge can be
thought of, to a rough approximation, as being the sum of a sphere
of positive charge and two oppositely directed electric dipoles, one
at the top and one at the bottom. That is, the only requirement for
an electric quadrupole moment is that the nucleus be squashed (or
stretched) along one axis.
When a nucleus possessing such an electric quadrupole moment is
subjected to an electric field that varies from place to place,
interesting things happen. The intrinsic electric quadrupole moment
of the nucleus and the electric-field gradient imposed from outside
together create distinct energy states. This result is analogous to
the multiple energy states in NMR, where the critical ingredients
were the intrinsic magnetic dipole moment of the nucleus and a
magnetic field imposed from the outside.
The key difference between NMR and NQR is the definition of
"outside." In NMR, the outside magnetic field arises
because the experimenter has invested considerable effort in setting
it up, perhaps using a superconducting electromagnet. In NQR, the
required electric field (or, more precisely, the required
electric-field gradient) comes for free: It reflects the local
arrangement of electrons around the nucleus under study. That
arrangement, in turn, depends not only on the nature of the atom but
also on its chemical environment. This feature accounts for one of
the chief benefits of NQR—the method is exquisitely sensitive
to chemistry.
Interestingly, an early motivation for investigating NQR was the
possibility that it might be useful for finding hidden explosives.
Shortly after World War II, Robert Pound, one of the pioneers of
NMR, became aware that people in the British army were speculating
about the possibility of using this technique to detect hidden land
mines. Pound was, however, skeptical that it would ever be possible
to project a magnetic field of the necessary uniformity into the
ground. So he decided to try NQR instead. As early as 1951, he
managed to produce some promising results, but for reasons that are
unclear, he did not pursue this avenue of research. A decade had to
pass before others began to appreciate the potential of this idea
and to study it in detail.
That later research has been carried out mostly in academic
laboratories in the United States and Europe, but NQR has attracted
military and commercial interest too. One of us (Miller) works at
the Naval Research Laboratory, where efforts to develop NQR for the
detection of explosives have been going on since 1987. The other
(Barrall) is employed at a private company, Quantum Magnetics, which
has been involved in similar efforts since 1993.
Explosive Mix
All of these efforts (going back as far as Pound's first tests) were
predicated on the realization that, because their chemical bonds are
somewhat unstable, nitrogen compounds are employed in virtually all
explosives. That use has a long history. Gunpowder, for example, was
first concocted some seven centuries ago from a mix of charcoal,
sulfur and potassium nitrate. The 19th century saw the introduction
of TNT—again another nitrogen compound: trinitrotoluene. And
such modern horrors as the truck bomb Timothy McVeigh used to blow
up the Alfred P. Murrah Federal Building in Oklahoma City contained
the fertilizer compound ammonium nitrate.
Thankfully (for our purposes), the nucleus of the common isotope of
nitrogen, 14N, is not spherical. It thus possesses an
appreciable electric quadrupole moment and can be detected using
NQR. Better yet, because the frequencies at which an NQR signal is
obtained reflect the chemical environment of the nitrogen nuclei,
one can distinguish dangerous explosive compounds from innocuous
materials that also happen to contain nitrogen.
The basic scheme for detecting hidden explosives is fundamentally
simple: One positions a loop antenna around a suspect suitcase or
over a patch of mine-infested ground and applies a short pulse of
radio-frequency magnetic field near the NQR frequency of interest,
which is usually something less than a few megahertz. The loop
antenna then serves to detect a faint return signal at the NQR
frequency if the material of interest is present in the vicinity.
One complication is that the strong outgoing pulse tends to set up
electrical reverberations in the antenna, just as banging on a bell
with a hammer sets up mechanical vibrations that can last a long
time. Although it might take only a few milliseconds for the
oscillations in a typical antenna to decay to negligible levels, the
return signal from some kinds of explosives lasts only a short time
too. The signal one gets back from TNT, for example, has a
characteristic decay time of less than one millisecond. So something
must be done to ensure that the left-over oscillations from the
transmitted pulse do not interfere with detection of the signal. A
similar concern arises in radar equipment, where the same antenna is
used to transmit powerful bursts of electromagnetic energy and to
receive weak echoes from distant objects. One solution (for both
radar and NQR) is to use special circuitry to dissipate the energy
left in the antenna right after the transmitted pulse is finished.
Another option for NQR is to use outgoing pulses that generate what
are called spin echoes.
Spin echoes are a phenomenon unique to nuclear resonance. Their
effect is to produce a measurable return from the nuclei under study
after the signal has nominally died out. How in the world can that
happen? The key is to understand that the reason the signal
disappears in the first place is not that the individual nuclei have
expended all the energy they have to give up. Rather, the overall
signal is lost because the separate emanations from individual
nuclei get out of synchrony. Spin echoes are induced using a
specially designed sequence of pulses, ones that coax the resonating
nuclei to come back into step at some later time.
A simple way to get the general idea is to imagine several runners
lined up at the beginning of a road race. When the gun goes off,
they all speed away from the starting line. Initially, it appears as
though the runners are advancing in unison. But because some go
slightly faster than others, after a short while they get out of
alignment. This is analogous to what happens in nuclear-resonance
experiments: The nuclei resonate at slightly different frequencies,
which causes their oscillations to drift out of phase, producing
little overall signal.
Now consider what would happen if the race officials instructed the
runners suddenly to turn around 180 degrees and head back to where
they started. At that moment, they would be at different places, but
(assuming that they all kept to their established paces) eventually
they would all arrive back at the starting line at the same time,
the slower ones having less far to go. In nuclear resonance, a
second pulse is used in essence to turn all the resonating nuclei
around so that at some later moment they all get back into phase and
produce a return signal that is well separated in time from the
outgoing pulse.
This tactic then helps to solve the ringing-bell problem. But there
is another fundamental concern in NQR: The signals are generally
quite weak. Indeed they are usually comparable in magnitude to the
noise that results from thermal agitation alone. So a considerable
effort has to be made to extract a reliable signal from background noise.
Because thermal noise arises in a completely random fashion, one can
boost an NQR signal simply by averaging the results over time or,
rather, over many repeated spin echoes. The NQR signal will increase
in approximate proportion to the number of spin echoes, whereas the
noise will rise only with the square root of that number. More
difficult is the problem presented by other forms of radio-frequency
interference, which could come from, say, distant AM radio stations
or from electronic equipment in the vicinity. In a controlled
environment (such as within a device for inspecting baggage), one
can employ suitable shielding, typically a grounded metal cage.
Dealing with such radio-frequency noise is, however, a greater
challenge for land-mine detection, where the space to be examined
cannot be enclosed. The solution adopted at Quantum Magnetics has
been to employ not one but several antennae. The additional
antennae, positioned remotely from the first, are used to record the
radio-frequency background at the moment the NQR measurements are
taken. This noise is then digitally subtracted from the signal
obtained from the main antenna.
Ground Truth
Although this technique shows a great deal of promise as a means for
finding hidden explosives, military tacticians consider NQR
detectors to be too slow to be useful in clearing roads of land
mines. In recognition of this concern, we are designing NQR
equipment to function as a "confirmation sensor." These
devices will supplement the conventional tools now being applied to
the task of finding land mines: metal detectors and
ground-penetrating radar, which can be quite sensitive but tend to
produce many false alarms. The idea is that an NQR sensor will be
used to test for the presence of explosives only at those spots
identified as suspicious by these other methods, reducing the number
of false alarms.
In an effort to gauge the effectiveness of NQR in this context, we
and some of our colleagues arranged in 2003 to test a prototype
confirmation sensor (built by Quantum Magnetics) under realistic
conditions. This equipment was designed to detect antitank and
antivehicle mines buried in roads. We performed these experiments at
two U.S. government test sites, one situated in the desert, the
other located in a temperate environment, so as to be able to gauge
whether damp soil, which can interfere with conventional detection
methods, posed special problems for NQR. (It didn't.)
We used a variety of mines for these trials: Some contained from 5
to 8 kilograms of TNT, whereas others used anywhere from 2 to 10
kilograms of an explosive called "Comp B," which is a
combination of TNT (40 percent) and the explosive compound
cyclotrimethylene trinitramine, better known as Royal Demolition
Explosive or RDX (60 percent). The mines were buried at realistic
depths, varying from 2.5 to 12.5 centimeters (as measured from the
surface of the ground to the top of the mine). These were blind
tests, in the sense that the people operating the NQR equipment did
not know ahead of time which of the hundreds of spots they examined
held mines.
We carried out the first set of trials at the desert site, both
during the day and at night. Why test day and night? Because we
anticipated that radio-frequency interference would pose a bigger
problem at night than during the day. (Recall how many more stations
your short-wave radio picks up after the sun goes down.)
Fortunately, the equipment dealt with this interference well, and
the results for day and night proved to be statistically identical:
The overall probability of detection was about 95 percent, and the
probability of false alarm was only between 4 and 7 percent.
In the second test at the temperate site, the TNT detection
probability was slightly reduced compared with what we had
determined under arid conditions. But we obtained similar results at
both sites for RDX. And again, the day and night tests gave nearly
identical outcomes: The overall probability of detection was once
more around 95 percent, and the probability of false alarms was
about 5 percent.
These tests clearly showed the feasibility of detecting antitank
land mines by NQR, but antipersonnel mines are a different matter.
Many antipersonnel land mines contain as little as 50 grams of
explosive, pushing current NQR detection sensitivity to its limits.
The 2003 tests made apparent some of the practical difficulties that
still limit NQR detection sensitivity. For the past few years, we
and our colleagues at the Naval Research Laboratory and at Quantum
Magnetics have worked (with support from the Army, the Marine Corps
and the Office of Naval Research) to overcome these problems with an
eye to developing rugged, portable hardware that can detect mines
swiftly and reliably under harsh field conditions. We've made
excellent progress in improving the sensitivity of our NQR
detectors, while at the same time making them more immune to
radio-frequency noise. These advances are bringing the NQR detection
of antipersonnel land mines into the realm of possibility.
Bad Bags
Of course, hidden explosives come in forms others than land mines.
Fortunately, one can design NQR detector coils to search for these
threats using fundamentally the same technology used to reveal land
mines. Instead of passing a small coil over a large patch of ground,
one typically moves a small (or perhaps not-so-small) object through
a large coil. Indeed, the coil can be quite sizable. Members of the
Defense Science and Technology Laboratory of the United Kingdom
recently constructed an NQR system for detecting ammonium
nitrate-based explosives hidden in the trunks of cars or the backs
of vans. They showed that a reasonable amount of the radio-frequency
field penetrates the vehicle, which is not too surprising when one
remembers that portable AM radios work just fine inside most cars.
They then demonstrated how a suspect vehicle could be driven into a
huge detector coil and rapidly scanned for the explosives.
Our initial efforts in NQR for the detection of explosives were
accelerated by the downing of Pan Am Flight 103 over Lockerbie,
Scotland, in 1988. Soon afterward, Miller and his colleagues at the
Naval Research Laboratory, with support from the FAA, began work on
an NQR system capable of scanning carry-on-sized baggage for the
presence of RDX-based explosives. This work showed that relatively
small quantities of this explosive (but enough to be a threat to
aircraft) could be detected in a reasonable time. Subsequently, the
Naval Research Laboratory licensed this technology to Quantum
Magnetics, which then built various prototype systems to scan larger
baggage for a range of explosive materials.
One component of NQR research at Quantum Magnetics has had the goal
of not only detecting the presence of an explosive substance but
also pinpointing its position within the suspect bag. It turns out
that localization, at least in one dimension, is easy enough to
accomplish. It requires only that many measurements be taken as the
bag passes along a conveyor belt through the NQR scanner. With these
observations, and knowing the physical characteristics of the coil
antenna used, one can readily calculate the position of an explosive
object (or objects) within the piece of luggage. Finding the
position of a problematic mass in two dimensions is not difficult
either: One needs only to rotate the bag 90 degrees and scan it
again. Indeed, a complete three-dimensional mapping can be
accomplished by rotating the bag a third time and scanning it once
more. (Actually, the results can be made more accurate by using a
larger number of scans, each one obtained after rotating the suspect
item into a different orientation.)
Of course, running a piece of luggage through a scanner many times
is bound to be tedious and time-consuming, particularly because the
operator would have to take care to adjust the orientation properly
during each pass. But the solution is straightforward: Run the bag
though once using multiple coil antennae of different orientations.
Quantum Magnetics has recently designed a system that employs two
perpendicular coils oriented at 45 degrees to the direction of the
conveyor belt. Although this arrangement does not allow the geometry
of an explosive to be mapped in any great detail, it does provide
two-dimensional localization in a single pass.
Beyond Explosives
Although the ubiquity of 14N in explosives makes NQR well
suited for detecting them, revealing hidden bombs is by no means the
only application of this technique. Narcotics, too, frequently
contain 14N, which opens the possibility of detecting
smuggled drugs of abuse. We have demonstrated detection of heroin
and cocaine in reasonable quantities with good sensitivity. However,
the great specificity of NQR, useful in differentiating explosives
and narcotics from other materials, can sometimes be a liability. In
particular, the detection of illicit drugs becomes rather
complicated because they exist in more than one form and because
their purity varies widely, causing NQR resonance frequencies to
shift and to broaden.
Although such changes are problematic for the detection of
narcotics, this phenomenon suggests another potential application of
NQR: for quality control in the chemical and pharmaceutical
industries. Work at the Naval Research Laboratory has shown, for
example, that the width of the NQR resonance lines in the explosive
RDX correlates with its sensitivity to detonation, an important
parameter in formulating explosives that are safe to handle.
In many crystalline substances, defects in the orderly packing of
atoms introduce strain at a microscopic scale, which in turn
influences the frequencies (and frequency ranges) of the NQR
resonance. Strain also can be induced by outside forces, and where
and how it builds up in structural materials can have especially
important consequences—namely mechanical failures. Not
surprisingly, a large sub-field in engineering is devoted to the
nondestructive evaluation of strain, an area in which NQR holds
great promise. For example, NQR may be especially valuable for
testing fiber-reinforced composite materials, which are found in
everything from tennis rackets to aerospace components. These
materials are not highly crystalline and usually do not contain a
significant number of quadrupolar nuclei, so they would not
typically provide an NQR signal.
This problem can be overcome in two ways: by embedding a small
amount of a crystalline substance containing quadrupolar nuclei
during manufacture of the composite material, or by later applying a
coating of such a substance to the finished structure. Tests on
fiberglass composites with embedded strain-sensing crystals,
performed last year at Quantum Magnetics, showed that NQR indeed
provides a very sensitive method of nondestructive evaluation.
The phenomenon of NQR allows, in principle, for even more ambitious
applications. For example, a number of research groups, including
those of Daniel J. Pusiol (National University of Córdoba in
Argentina), Rainer Kimmich (Ulm University) and Bryan H. Suits
(Michigan Technological University), have demonstrated the potential
for NQR imaging and for the spatial localization of strain. These
early efforts suggest that it may one day be possible to obtain
high-resolution images showing the distribution of everything from
temperature and strain state to chemical composition and purity.
Given the rapidity with which MRI moved out of the laboratory and
into hospitals, it seems fair to wonder: Will the benefits of NQR
prove great enough to spur similarly dramatic advances?
Bibliography
- Barrall, G. A., L. J. Burnett, K. A. Derby, A. J. Drew,
K. V. Ermolaev, S. Huo, D. K. Lathrop, T. R. Petrov, M. J.
Steiger, S. H. Stewart and P. J. Turner. 2005. Nuclear
quadrupole resonance for landmine detection. Proceedings of
the Sixth Joint International Military Sensing
Symposium (in press).
- Barras, J., M. J. Gaskell, N. Hunt, R. I. Jenkinson, K.
R. Mann, D. A. G. Pedder, G. N. Shilstone and J. A. S. Smith.
2004. Detection of ammonium nitrate inside vehicles by nuclear
quadrupole resonance. Applied Magnetic Resonance 25:411–438.
- Garroway, A. N., M. L. Buess, J. P. Yesinow-ski, J. B.
Miller and R. A. Krauss. 1994. Explosives detection by nuclear
quadrupole resonance (NQR). Proceedings of SPIE 2276:139–149.
- Garroway, A. N., M. L. Buess, J. B. Miller, B. H. Suits,
A. D. Hibbs, G. A. Barrall, R. Matthews and L. J. Burnett. 2001.
Remote sensing by nuclear quadrupole resonance. IEEE
Transactions on Geoscience and Remote Sensing 39:1108–1118.
- Miller, J. B. 1998. NMR imaging of materials.
Progress in Nuclear Magnetic Resonance Spectroscopy 33:273–308.
- Robert, H. and P. J. Prado. 2004. Threat localization in
QR explosives detection systems. Applied Magnetic
Resonance 25:395–410.
- Vierkötter, S. A., C. R. Ward, D. M. Gregory,
S. M. Menon and D. P. Roach. 2003. NDE of composites via
quadrupole resonance spectroscopy. Proceedings of SPIE 5046:176–184.